Distributor Making Adventures

[petertha]

Introduction
I thought I’d post some pictures of my bench distributor work thus far. By bench I mean it’s a generic tester not associated to a specific engine. I’m trying to figure things out & make sparkies on a row of RCEXL plugs by finger spinning the shaft. Hopefully I will be able to integrate learnings into the next REAL spark engine.

I leveraged designs & ideas from others, so nothing dramatically different in that regard. Maybe what is a bit unique, or less prevalent, is how I went about the distributor cap and other parts using 3DP & molds. So that’s mostly what I want to show here. Anyway, thanks to some fellow ME’s who answered questions or shared their knowledge: George Britnell, Steve Huck, Terry Mayhugh, Mike Tull, Ray Dean, Dave Sage, Dan Williams, Charlie Welkie…probably forgetting some.

My 5-cyl radial used glow plugs which was integral to the design. I wanted it to run so played it safe, plus I had some prior RC experience. Glow ignition is comparatively simple but also pretty much confined to methanol fuel, needs specific compression ratios, has no real timing control or advance/retard. Miniature spark ignition seemed like a distant, intimidating pipe dream; something I knew very little about.

My test distributor body is ~24mm OD, the cap is 26mm OD x 21mm tall. My goal was to increase the terminal density up to 8 or 9 cylinders on the same platform to see if and when I’d encounter problems. I don’t know if this can really be determined with certainty without a scope or having running on the engine.

[petertha]

Signaling
Although I’ve seen some incredible examples of builders who stuffed mechanical points inside a 1” distributor body, I focused on the Hall sensor type. There are 2 flavors I’m aware of. (1) a magnet wheel style which has N embedded magnets for N cylinders. When the wheel rotates over the sensor, each magnet fly-over signals the ignition module to fire.

(2) a shutter plate or drum with N window openings for N cylinders spinning between a single, stationary magnet positioned opposite the sensor. Each opening allows the sensor to see the magnet, otherwise it is blocked by the solid element. My understanding is this style can provide a cleaner, more reliable signal with less proximity magnetic noise. The open width corresponds to dwell time which can factor into coil-based ignition systems. A potential downside is the shutter wheel starts to look like Swiss cheese as the cylinder count increases, particularly within a small body diameter. But it has been done.

Some screen grabs from this Allegro link which is quite good reading
https://www.allegromicro.com/en/insights-and-innovations/technical-documents/hall-effect-sensor-ic-publications/hall-effect-ic-applications-guide

The reason I mention this is because I was hoping to use the same or very similar distributor body to try both mag wheel & shutter methods.

[petertha]

Design
Here is an assembly section markup showing the bits & pieces. Not everything is accurate in these views as I was changing things on the fly. I draw in metric because of hardware sourcing, but I convert & work in Imperial on the machines.

[petertha]

Distributor Cap
Most shop made distributor caps I’ve come across were machined from plastics like Delrin, occasionally reinforced epoxy/glass. Some kind of machinable, dielectric material.

The round body & internal cavity would be straightforward turning on a lathe, but other features are involved. Cutting out the terminal towers requires something like an annular cutter tool. Real distributor towers usually have fillets around their bases & can be tapered from top to bottom. Or they can be set slightly outside the cup OD. The cap needs to be retained to the body so requires something like bolt lugs or side clips. Many of these cosmetic features don’t make it perform any better so could be simplified or omitted. CNC would be the weapon of choice if available. Anyway, capturing some of these FS features was my motivation logic to try using 3DP & mold making methods. I’m relatively new to 3DP but familiar with 3D CAD. I have some model composite experience, but not really this particular stuff.

[petertha]

I designed a generic cap with the idea I could just specify N terminals in CAD & the model would self-generate another configuration which would fit the same distributor body. My first attempt was to make a male master part from aluminum thinking it would be dimensionally accurate & nice surface finish etc. It had tapered terminals set into top holes. But it was a lot of work for something that was likely going to change. And I would still have had to Bondo some fillets. 

So, I went the 3DP route & printed a male master part (aka ‘plug’) using PLA. I used a 0.4mm nozzle which made a reasonably accurate part, but the print striations around the fillets & seams were noticeable. Hiding & blending involved some mini-autobody finishing – a few cycles of priming & sanding. Nothing too crazy but trying to make it look presentable. Alternating primer colors & mini sanding tools helps. On later designs I switched to a 0.2mm nozzle & it was noticeably better quality requiring less prep work. I tried spraying a coat of paint hoping for a uniform, shiny finish, but the results were not great. It was hard an get an even coat without blind spots behind the towers or inadvertent sags & bloops due to the finicky shape. So, I left it in primer state with no adverse molding issues.

I learned that it’s beneficial to add ~2mm thickness to the plug’s entire base footprint which of course is reproduced in the female mold. It provides fill leeway for the urethane pour. If the plug is made to exact height, it’s difficult to fill the corresponding female mold to the brim without spill-over & leaves little wiggle room for machining to dimension. The extra 2mm fill helps & just means machining a bit of extra material off.

[petertha]

The finished plug sits on a 3DP base plate, centered on a 3mm pin. A 3DP mold dam fits over the plate. The inside dam corners are chamfered to save mold material which doesn’t really contribute to anything. I made one corner a slightly different size to act as a key register so the mold is uniquely orientated if removed. A releasing agent was sprayed on the assembly. I used Ease-Release-200 for everything.
https://sculpturesupply.com/products/ease-release-200?_pos=1&_psq=ease&_ss=e&_v=1.0

The silicone was mixed & then de-gassed in a vacuum pot. The mixture foams & rises as entrained air bubbles expand & escape. Then it collapses on itself leaving the mix mostly bubble free which is the objective. Degassing only takes about a minute but helps make a void free mold surface if all goes well.
The silicone is slowly poured into the cavity from one corner, allowing it to flow out. Once cured (overnight) it becomes the female mold for casting the distributor part from a chosen resin. The mold is semi flexible, about the durometer of a soft eraser for reference. It can be re-used multiple times. The mold should stay resident in the 3DP dam when casting the resin. That maintains dimensional stability as much as possible, but still allows the part to be easily demolded.

The shrinkage % of silicone is amazingly small but it’s all relative to the application. In my case even if the cap diameter was out 0.005-0.010” it’s not really the end of the world because the critical dimensions are machined from the solid cast urethane part anyway. YouTube is filled with interesting videos of very complicated parts cast from silicone molds, many with severe undercuts. These would be very difficult if not impossible with ‘hard’ molds. 

[petertha]

Silicone
I tried 2 different brands of silicone. Links attached so you can read the specs. Smooth-On Moldstar 40-T which is a milky clear color.
https://sculpturesupply.com/products/mold-star-40t?_pos=5&_fid=dbb982474&_ss=c

A product from China available on Amazon called BBDINO 30A which comes colored blue.
https://www.amazon.ca/dp/B08CVSD6MZ?ref_=ppx_hzsearch_conn_dt_b_fed_asin_title_3&th=1
The BBDINO was less expensive (to me). Both worked well enough for this purpose.

Both are 1:1 mix by volume. 40 & 30 refer to the corresponding durometer ratings. I wanted it to be high enough that the mold would not distort much, but flexible enough to de-mold the finished part. It seems like durometers above 40 in silicone are less common, at least in small volumes. I also wanted to use the same stuff to make other kinds of molds as well as ignition related parts like spark plug boots & seals which might favor a lower durometer. Anyway, both products seemed to work for the cap, but I kind of favor the 40. BBDINO claims that no degassing is required but I don’t quite agree with that, or maybe its part specific. It foamed madly as well in the vacuum chamber. BTW silicone can be colored with dedicated pigments.

Soft durometer urethanes are also available & can generally be used in the same manner as silicones. I intend to experiment one day soon. But urethanes have a much shorter shelf life, even when using a dedicated gas blanket extender. But they also come in harder durometers which may be an advantage in certain applications.

[petertha]

Urethane
The distributor cap was made by casting urethane resin into the silicone mold. I used Smooth-Cast 300 which becomes a hard plastic ~70 D Shore when cured. It is an easy 1:1 mix by volume but you have to be pretty exact. I don’t think I ever figured out dielectric properties beyond anecdotal reference it should be similar to other plastics. The spec sheet says 50 deg-C heat deflection temp which should be OK.

https://sculpturesupply.com/products/smooth-cast-300
https://www.sculpturesupply.info/pdfs/technicalbulletin/smoothcast300series.pdf

This particular resin has a pretty short pot life, 3-4 minutes, which means only about a minute of mixing time & you need to be pouring. Fortunately, the cap volume is small so not a big issue. I’m going to try 305 next which has a slightly longer pot life. This resin mix is only 80 cp viscosity which is quite thin & a desirable property because it flows nicely into cavities & better allows entrained air bubbles to escape on their own. Voids are still possible depending on the mold shape, or if you pour too quick. Urethane could be degassed in a vacuum pot but you are really working against the short pot life clock, it’s probably impractical. What I see more often is after pouring, the mold is transferred into a pressure chamber which squishes micro bubbles to where they are not noticeable.

Demolding is very easy. You just squeeze around with your fingers & out it comes. The urethane cures white but can be colored opaque with their specific line of pigments, which is how I did the red & black ones. I’ve read that black plastics may contain carbon in the pigment & might adversely conduct electricity. It was hard to be certain but the black pigmented cap seemed to spark the same as others so maybe different black colorant or concentration?

I also tried rattle can painting some caps. It seems to have bonded & cured so maybe another option. You absolutely have to scrub any release agent residue with soapy water & paint prep or the paint will fisheye.

I also cast one cap from epoxy resin with some milled glass fibers added as a test & it turned out OK, so I’m confident that the same ‘soft mold’ technique could be used. Epoxy is typically a higher viscosity, especially if its more paste like with additives, which means higher probability of entrained air & void potential. OTOH, epoxy usually has longer pot life, so maybe could be put in the vacuum pot during cure just like regular composites work. A benefit of this kind of mold is making multiple parts if desired.

So why cast a completely solid distributor cap body versus something more like a finished cap with internal cavity? Because it much easier & the resin waste is negligible. A solid allowed me to machine different ceiling heights (counterbore depths) corresponding to different rotor sizes & the skirt lip feature. I think it would be finicky to pour into a finished dimension shell mold with only 2mm wall thickness, the mold would be more complicated.

I did some tests on cured urethane samples by drilling holes & bonding metal like brass, mimicking terminals in the towers. I was satisfied that ordinary CA & epoxy seems to hold it quite well as long as it’s a sanded or machined surface. When I tried this same test on some Delrin, I really struggled to make it stick which seems consistent to what I’ve read. Apparently, there are specialized glues & primers but expensive. I think for Delrin the parts need to have some kind of knurling texture, be more interference fit or somehow mechanically keyed.

[petertha]

Machining
The urethane part cures hard after about 10 minutes & can be demolded, but I usually gave it a couple hours to stabilize before machining. To machine I held the round part of body in a chuck or collet, then skimmed the base off flat to dimension in the lathe. Urethane machines exceptionally well.

One could either machine the internal cavity first & glue in the terminals second. I initially did this so that the terminals extend ~1mm proud of the cap ceiling. To position I first put a few layers of tape over the rotor mimicking the desired air gap & pushed the terminals down until they contacted the tape. Alternatively, one can glue the terminals into tower holes beforehand so that machining the cavity trims both the urethane & terminal ends simultaneously in one go. This would make the ends flush with the ceiling or making the corner birds beak profile or whatever it’s called

[petertha]

Terminals
I used 2mm brass banana plug connectors sourced from RC model hardware. The fit is reasonably tight. They have pretty low resistance, nominally rated for 20-30 amps continuous. 16 AWG spark plug wires were soldered to the male plug end which has a little solder pot feature. The matching tubes become the terminal contact once glued into the cap. The tubes are 2.5mm OD x 2mm ID x ~15mm length. A bit longer than I would have preferred, but they worked out OK. You just need to be careful trimming what will become the terminal ends inside the cap without cutting through & exposing the tube portion, not a lot of wiggle room here. I tried a few drill jig styles to hold & orient the cap. 3DP to the rescue again. The 2 mounting lugs were drilled M1.6 clearance using the same jig.

On another cap I used brass M3 set screw glued into tapped holes. They were landed proud of the terminal top to accommodate a nut, so spark plug wires come in laterally & attach to the set screw via a M3 ring terminal. The cap terminal towers could be trimmed shorter in this mode if desired. One could also make your own terminals & center tap a hole. Many possibilities on the same basic design.

[petertha]

Distributor Body
The body was turned from 6061-T6. The shaft is 4mm OD, some kind of ground stainless stock from AliExpress. It spins on 2 bearing races, 8mm OD x 4mm thick. First the slimmer neck segment was turned & drilled, then part flipped to turn the cup features. Then over to the RT to cut away the lip material for the 2 mounting lugs. Then into the mill vise to machine the hall sensor features using the lugs as reference.

The plan is for the distributor neck to reside within a bushed hole on the engine block & have some kind of arm to rotate it for advance / retard position, somehow linked to the throttle.

[petertha]

More pics

[petertha]

Magnet Wheel
Made from 6061-T6. I guess it could be made from plastic type materials, but the same comment applies as the rotor – reliable set screw threads & adhesive to hold the magnets in position.
The lower hub accommodates a M2.5 stainless set screw to fix it to the shaft. I chose SS because it’s supposedly non (well, ‘much less’) magnetic vs a plain steel which might become magnetized & trick the sensor. The plate boss is the same 2mm thickness as the magnets which come in 2 common metric sizes: 3mm & 2mm diameter. After confirming polarity with a cheapo compass (south pole attracts the north seeking needle) they were retained with CA glue. You could probably use Loctite too.

Initially tested some low-cost, craft grade magnets because they were readily available. They seemed to signal fine until the gap becomes excessive. From memory they got fainter ~5mm gap on the 2mm diameter & maybe 8mm on 3mm diameter. Magnetic strength increases with thickness, but it’s not proportional & I was trying to keep the wheel compact. My preference was 2mm diameter because I could space 8 on a 20mm disc (16mm mag circle) with seemingly consistent signaling. 3mm diameter gets crowded but I think 4-6 magnets might be OK. 8-cylinder might be getting busy on a distributor this size for other reasons anyway as the terminals get crowded too. Magnet/sensor distance is a deeper subject that involves other variables but because I was able to get them spaced 1-2mm, I figure 2mm diameter magnet would be a safe bet. Also, I found a source of N52 which should be stronger & hopefully more consistent quality, so TBC.

I made provisions for a 0.5mm thick shim washer between the underside of hub & bearing ID so it doesn’t rub on the OD. This could have been machined in the hub but a shim washer also gives some wiggle room to displace the mag wheel / rotor stack up or down a bit accounting for machining tolerances.

[petertha]

Sensor Holder
Sensor dimensions vary a bit by manufacturer but are nominally 4x2x3mm. Three wires extend from the sensor block which are soldered to a typical RC 3-wire lead that plugs into the ignition module. I used Honeywell SS441A & Allegro A1104LUA-T, but these PN’s relate to the ignition modules themselves. The magnet faces the beveled face, not the end or the larger face.

Of note, I was made aware of this sensor tester gadget which came in handy. It is powered by a button cell. You just plug it into the RC sensor lead, the LED and buzzer come on when the magnet activates the sensor. So quite useful for checking & setting timing. This one came from AliExpress but they might be available elsewhere where RC ignition systems are sold. I don’t think it’s designed to indicate relative signal strength but the buzzer & LED seems to fade & then nothing as the distance is progressively increased. Electronics savvy people probably have more sophisticated instrumentation but for me this was better than nothing.

[petertha]

I wanted a way to mount the sensor into the distributor body with some consistency so it offset the magnet face by ~ 1-2mm & was also centered radially relative to the magnet diameter. That could have been as simple as gluing it to the distributor floor, which would be the most compact arrangement. Maybe not as well protected or electrically insulated or removable when the time came.

Design 1 was a 3DP plate attached to the floor under the mag wheel which gave it needed elevation anyway. It has a recess pocket for the sensor which I intended to glue in place. This worked perfectly fine, but I didn’t consider that unbolting the plate doesn’t do much good because I could not feed the harness plug through the side slot opening anyway. So, if the sensor died, it would be the same snip the cable routine. The 3DP part is a cheap consumable.